0
\$\begingroup\$

I can't seem to find a coherent and thorough answer to this question anywhere on the internet. I know about this Stack Exchange post asking a similar question in the title but none of the answers really answer the question in the title, rather answering very specific questions asked in the body. I actually want to know exactly this:

Which electrical components can be (reasonably) accurately tested/measured while still soldered to the board and under which circuit circumstances?

An exhaustive list would be deeply appreciated so that there will hopefully be this one place people can find the answer to this question for all common components.

I know that it's ideal to remove components before measuring but much of the time that's not feasible. I know that there are some circuit circumstances in which you could get a good reading for a component and others where you can't so that basically has to be included in the question. Also, I know that the accuracy of basically any measurement is going to be lessened slightly by the component being connected to the circuit, I'm not talking about 100% accuracy, just to be able to get a reasonably accurate measurement to determine if the component is dead or way out of spec. Lastly, it's OK if the measurement would require some special instrument, for example I know that capacitors can be tested while in-place using an ESR meter.

\$\endgroup\$
10
  • \$\begingroup\$ why do you require an exhaustive list? \$\endgroup\$
    – jsotola
    Feb 9, 2023 at 6:54
  • \$\begingroup\$ please focus your post to one question \$\endgroup\$
    – jsotola
    Feb 9, 2023 at 6:55
  • \$\begingroup\$ I want an exhaustive list so that there will hopefully be one single place that answers this question for all components, or at least all commonly used components. I will focus the post to just one question \$\endgroup\$ Feb 9, 2023 at 7:06
  • 1
    \$\begingroup\$ Thats tough to answer, because it really depends on the circuit. Some sensitive parts might even be damaged if you probe around in the wrong connections (although most modern ics should withstand sth like that). But even with "dumb" circuits where you only have resistor networks you need to know the schematic - a 10k resistor might read 5k but be completely fine, because there is another 10k somewhere in parallel. Dead shorts are easy to spot (e.g. a 10 V rail that has 2 Ohms to GND) but thats often not easy to localize. My suggestion to you: read up on methodical fault finding in circuits.. \$\endgroup\$ Feb 9, 2023 at 7:09
  • 1
    \$\begingroup\$ There can be no answer to that, because if you have a component on board, it depends how it connects to other components. So you need to know how each component is connected, to be able to make a decision if you can reasonably accurately measure it, or not. Also, you can't expect to measure capacitors in-circuit with an ESR tester, unless you know there are no circuits attached to the capacitors that can affect the measurement. So your assumption is incorrect. \$\endgroup\$
    – Justme
    Feb 9, 2023 at 7:24

4 Answers 4

5
\$\begingroup\$

The main problem is when the component is in parallel with other components. Maybe the simplest example of the problem is when you try to measure the capacitance of a decoupling capacitor from VCC to GND. There could be dozens or hundreds of capacitors in parallel. When you try to measure one of them, you are actually measuring all of them.

But that is not the only problem. If you have, for example, a pulldown resistor on a GPIO pin, when you try to measure the resistance of the pulldown resistor, some current MAY flow through the processor. So if the resistor is supposed to be 10 k, it may measure out at 10 k or any lower number depending on how much current flows through the processor.

Generally, the circumstance when you can measure a component in circuit is when the component is not in parallel with any other component. Then you can measure it accurately in circuit. So it is not about what type of component you are measuring, but what else is in parallel with it.

I am afraid that a truly comprehensive treatment of the issue would be too long for this format. What you should take away from this is that it is OK to try measuring a component, but then you have to consider the whole circuit and how it affects the measurement.

Also, measurements can be used as kind of a fingerprint. "Bad" boards will often measure differently than "good" boards. So even though the measurement may not correspond to the real component value, it can still be a useful diagnostic.

\$\endgroup\$
5
  • \$\begingroup\$ Thank you. In your pulldown resistor example, would that still be true even if you had the leads of a multimeter directly connected to either side of the resistor (without any of the processor between the leads (ignoring parasitic resistance / lead resistance)? Also, the tip about comparing a good board and a bad board as relative references is super valuable so thank you \$\endgroup\$ Feb 9, 2023 at 7:46
  • \$\begingroup\$ Also to clarify, when you say parallel components, does that only mean if both parallel components are between the meter's leads? In other words, if we have two parallel resistors in the circuit, if I put the leads around the legs of just one of them, will it still be thrown off by the other one? Sorry if this is an obvious question, I think I know the answer but I wanted to ask it here to make sure and so that others can see \$\endgroup\$ Feb 9, 2023 at 7:55
  • \$\begingroup\$ Lastly, what about making other measurements, like using a voltmeter or an ammeter? Are those also mostly only thrown off by the existence of parallel components, or is it a totally different story? \$\endgroup\$ Feb 9, 2023 at 7:56
  • \$\begingroup\$ @SamwiseGanges I adressed your comments in another answer. I didnt want to edit mkeith's since I think it is a very good answer to the original post, adding more would probably lead to confusion. \$\endgroup\$ Feb 9, 2023 at 8:39
  • \$\begingroup\$ To answer my own questions above, I should have realized that there is no difference between putting the meter leads on either side of a parallel circuit and putting the leads on either pin of a single resistor which is in a parallel circuit. Electrically they are identical, it just seems at first glance that the second option is isolating the resistor. In both of them, the current will flow out the positive lead, through both components, then back through the negative lead. You always need to simplify the circuit diagram \$\endgroup\$ Feb 9, 2023 at 21:33
2
\$\begingroup\$

This answer is addressed at OP's comments left on mkeith's answer (too large for a simple comment reply)

RE Pulldown confusion: "Leads directly at the resistor" doesn't make any difference.

schematic

simulate this circuit – Schematic created using CircuitLab

I couldnt find a resistor measurement symbol so I used arrows here. If you measure the resistance and put your probes at "probe pair 1" it will measure exactly the same as at "probe pair 2". Think of it this way: When you measure on a PCB, you dont get the value of a specific part, but for everything between your leads. It doesnt matter where you put your probes along the way, since the controller will still be in parallel to the resistor.

RE parallel components: Here is an exact replica of a repair job at my work, where one of the new guys was confused about the measurement.

schematic

simulate this circuit

He wanted to verify the ratio of the voltage divider on the right, but when he put the probes at the marked place to get the value of R1, he measured 7.5k. After removing it, it suddenly measured 10k. Reason: On another page in the schematic, we found the 5 V regulator, that generates the 5 V via a feedback network. And since both circuits are connected (all "GND" are connected with each other, same with "5V"), R1 is essentially in parallel with 3 other 10k resistors.

RE Voltmeter/Ammeter: Important disclaimer: If you make resistance measurements, make sure the board is not powered! Otherwise, you will destroy your multimeter.

Current measurement: When searching for faults it might be helpful to know the current draw of the circuit, so yes, in a specific case that makes sense. But for that you need a supply that displays the current drawn - and it is only a general hint that something is wrong (e.g.: a board that usually draws 100 mA draws 500 mA now). I can't think of any other example where measuring current would be helpful.

Voltage measurement: That is mostly used to find out if a specific signal is low or high (e.g., to find out if a uC correctly sets an output) or if a supply voltage rail is present (measuring 7 V on a 24 V rail might hint at a partial short between the rail and another net).

Most multimeters also feature a diode mode, that can be helpful as well.

\$\endgroup\$
2
  • 1
    \$\begingroup\$ @Downvoter please explain why, I'd really like to know if there is a specific part to the answer that is wrong. \$\endgroup\$ Feb 9, 2023 at 9:23
  • \$\begingroup\$ Thanks a ton, yes I realized my stupidity earlier today when talking to my Dad, of course it doesn't matter where the leads are on a parallel circuit, it's electrically identical because there is a junction on either side so the current will flow through both components. All very useful information thanks! \$\endgroup\$ Feb 9, 2023 at 21:41
0
\$\begingroup\$

Unless you know exactly where in the circuit the component is and can be sure the circuit will not provide any other component in series or parallel the answer is none.

\$\endgroup\$
1
  • 2
    \$\begingroup\$ It is fine to have something in series. Just not in parallel. \$\endgroup\$
    – user57037
    Feb 9, 2023 at 7:20
-1
\$\begingroup\$

This is a difficult problem. Components tend to be connected on a board, and these connections corrupt the readings.

Many manufacturers, starting in the 1960s when automatic test equipment was just starting to be computer controlled, like Teradyne and Marconi Instruments to mention a couple, came up with a way of making some of these measurements.

In the best case, these connections can be 'guarded out'. This is essentially an extension of a 4 wire measurement.

Let's say you have a pi attenuator R1/2/3, and you want to measure the series resistor R2. If you simply put a resistance meter across the terminals, you would also see the effect of R1 and R3 forming a parallel path.

schematic

simulate this circuit – Schematic created using CircuitLab

To make a guarded measurement, you would identify a terminal on each parallel path. These are the guards. In this case, the guard terminal is the board ground. Using the sink amplifier to hold the far end of R2 at measurement ground, use the source amplifier to apply a known voltage across R2. The R2 current flows into the virtual ground of the sink transimpedance amplifier, and is measured. This allows you to calculate R2.

Some current flows through R1. This does not affect the source amplifier output voltage, so causes no error. The voltage across R3 is held at zero, so no current flows through it, and it causes no error.

This technique will work with DC or AC.

There are limitations to this technique. Sometimes you can't find a suitable guard terminal, two components may actually be in parallel.

Errors in the amplifiers limit the dynamic range that can be achieved. An offset error in the sink amplifier results in some voltage across R3. If R2 and R3 are similar sizes, then this does not degrade accuracy too much. If R3 is very small compared to R2, then it does. It's common to have inductors on a board, which will have very low DC resistances, and cause this dynamic range problem to surrounding resistors.

\$\endgroup\$

Not the answer you're looking for? Browse other questions tagged or ask your own question.